AI-Powered Identity Verification & Fraud Detection

About the project

Developing an AI-powered identity verification system for Hamqadam that automates the verification of user identities using CNIC documents and selfie-based facial analysis. The system combines computer vision, OCR, face recognition, duplicate detection, and risk-based decisioning to determine whether an identity can be automatically verified or requires human review.The pipeline first validates uploaded image quality, extracts identity information from the CNIC using OCR, and compares the user’s selfie against the photograph on the identity document. It also performs duplicate-face checks to identify repeated registrations and generates a fraud risk score used to route cases toward approval, rejection, or manual review.FastAPI services expose the AI pipeline for integration with the main Hamqadam platform, allowing straightforward cases to be processed automatically while directing uncertain cases to reviewers. The system is currently under active development.